Four incredible exhibitions
Exhibition visitor badge is free
>
25 - 30 January 2025
San Francisco, California, US
Conference 13293 > Paper 13293-9
Paper 13293-9

Deep learning-based detection of prostate cancer grades and chronic prostatitis in biparametric MRI

25 January 2025 • 1:30 PM - 1:50 PM PST | Moscone South, Room 301 (Level 3)

Abstract

Biparametric MRI (bpMRI) has become an essential tool for non-invasive prostate cancer (PCa) screening. This study aims to develop and evaluate a deep learning-based computer-aided diagnosis (CAD) system for automated characterization of PCa grades and chronic prostatitis on bpMRI. We collected 1647 biopsy-confirmed findings, including Gleason scores and prostatitis, from 1074 patients. A 3D convolutional neural network based on the previously established nnU-Net architecture was trained on T2-weighted and diffusion-weighted imaging (DWI) sequences. The best-performing model configuration, using fine-grained class granularity (including prostatitis) achieved a lesion-wise partial Free-Response Receiver Operating Characteristic (FROC) area under the curve (AUC) of 1.94 (95% CI: 1.76 to 2.11) and a patient-wise ROC AUC of 0.874 (95% CI: 0.793 to 0.938) for clinically significant PCa detection. Inclusion of prostatitis as an auxiliary class improved specificity by 3 to 7% across different class granularities. (This study demonstrates that a deep learning-based Computer-Aided Design (CAD) system can effectively detect and grade prostate cancer while simultaneously identifying chronic prostatitis on bpMRI. The ability to detect benign conditions alongside cancer grades may improve the interpretability and clinical utility of CAD systems for prostate MRI analysis.

Presenter

Lucas Engelage
Laser-Forschungslabor, Klinikum der Univ. München, Ludwig-Maximilians-Univ. München (Germany), ALTA Klinik GmbH (Germany)
Lucas Engelage is a dedicated researcher with 4 years of experience in the field of urological diagnostics and treatment. His work focuses on advancing prostate cancer detection and management through innovative technologies. Engelage has been at the forefront of developing an artificial intelligence system capable of detecting prostate cancer in MRI images, significantly improving diagnostic accuracy and efficiency. In parallel, he has been instrumental in the development of a novel PSA test that utilizes dry capillary blood, potentially revolutionizing the ease and accessibility of prostate cancer screening. His research also extends to therapeutic innovations, including work on the TULSA PRO therapy, a minimally invasive treatment for prostate cancer. Engelage's multifaceted approach to urological research demonstrates his commitment to improving patient outcomes through the integration of cutting-edge technologies in diagnostics and treatment.
Application tracks: AI/ML
Presenter/Author
Lucas Engelage
Laser-Forschungslabor, Klinikum der Univ. München, Ludwig-Maximilians-Univ. München (Germany), ALTA Klinik GmbH (Germany)
Author
ALTA Klinik GmbH (Germany)
Author
Otto-von-Guericke Univ. Magdeburg (Germany)
Author
ALTA Klinik GmbH (Germany)
Author
ALTA Klinik GmbH (Germany)
Author
ALTA Klinik GmbH (Germany)
Author
ALTA Klinik GmbH (Germany)
Author
ALTA Klinik GmbH (Germany)
Author
ALTA Klinik GmbH (Germany), Laser-Forschungslabor, Klinikum der Univ. München, Ludwig-Maximilians-Univ. München (Germany)
Author
Laser-Forschungslabor, Klinikum der Univ. München, Ludwig-Maximilians-Univ. München (Germany)